Multi-output Ensembles for Multi-step Forecasting

TL;DR

This paper investigates the use of dynamic multi-output ensemble methods for multi-step ahead time series forecasting, revealing their strengths and limitations compared to static ensembles across different forecast horizons.

Contribution

It provides a comprehensive analysis of dynamic ensemble strategies for multi-step forecasting, highlighting their performance and challenges over static methods.

Findings

Dynamic ensembles with arbitrating and windowing perform best overall.

Most approaches struggle to outperform static ensembles as forecast horizon increases.

Experiments conducted on a large dataset of 3568 time series.

Abstract

This paper studies the application of ensembles composed of multi-output models for multi-step ahead forecasting problems. Dynamic ensembles have been commonly used for forecasting. However, these are typically designed for one-step-ahead tasks. On the other hand, the literature regarding the application of dynamic ensembles for multi-step ahead forecasting is scarce. Moreover, it is not clear how the combination rule is applied across the forecasting horizon. We carried out extensive experiments to analyze the application of dynamic ensembles for multi-step forecasting. We resorted to a case study with 3568 time series and an ensemble of 30 multi-output models. We discovered that dynamic ensembles based on arbitrating and windowing present the best performance according to average rank. Moreover, as the horizon increases, most approaches struggle to outperform a static ensemble that assigns equal weights to all models. The experiments are publicly available in a repository.